Beam Monte Carlo official figures from the January 2000 meeting.

      These 3 figures were agreed upon in the collaboration as 
acceptable for official plots. 

Figure 1:
  Nuclear model effects on final answers.
  The version shown in the collaboration meeting had as its X-axis
  the 'extrapolation ratio' for a specific set of inputs (water target,
  2 m integration region). This version plots the exact same thing, but 
  relative to the value actually used in a full analysis. That is:
  X value now = 100% * ( (X value previous/ X central value) - 1. )
  It shows explicitly how we arrive at the +5-6% error term for 
  nuclear model dependence; just look at the range this covers.
  Ask JH for related figures about how the models were derived. 
  Those are not official figures, but useful for collaborators to
  understand if they want to present this official one.

Figure 2:
  Compilation of known systematic errors. (Unchanged since meeting)
  This list of errors was the main topic of the whole talk at the meeting.
  It is given as an estimate of the error on final expectation at SK
  for an analysis that is based on Monte Carlo for extrapolation.
  This is separate from error on any analysis that relies on the 
  pion monitor for extrapolation, although the final numbers happen to
  be about the same. 

Figure 3: 
  Comparison of various models to pion monitor data.
  In this figure (which is a slight update from that shown at the meeting)
  a Chi^2 is calculated for the shape of the pion spectrum from each
  of 64 'alternate nuclear models' compared to the fitted pion spectrum
  made by Maruyama_san from pion monitor measurements.
   The updates since the meeting version are:
   - Only the spectrum (no angular information) is used.
     The removes effects of any systematics of angular determination in the fit.
   - The word "Fitting" is changed to "Comparing" because this is
     more accurate. Each MC spectrum of pions is fit for a normalization,
     but the resulting reduced Chi^2 values are not used in any minimization,
     so it's not really a fit.
   - The values for GCalor(GFluka) and our standard MC model are explicitly
     put on the figure.
   * NOTE: We should not use the word "disfavored" about GFluka based
     on this. (This was the result of some long discussion at the meeting.)